sglang_v0.5.2/sglang/test/srt/hicache/test_hicache_mla.py

68 lines
1.7 KiB
Python

import unittest
from types import SimpleNamespace
from sglang.srt.utils import is_hip, kill_process_tree
from sglang.test.run_eval import run_eval
from sglang.test.test_utils import (
DEFAULT_MLA_MODEL_NAME_FOR_TEST,
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
DEFAULT_URL_FOR_TEST,
CustomTestCase,
popen_launch_server,
)
_is_hip = is_hip()
if _is_hip:
hicache_args = ["--hicache-size", 200]
else:
hicache_args = ["--hicache-ratio", 2]
class TestHierarchicalMLA(CustomTestCase):
@classmethod
def setUpClass(cls):
cls.model = DEFAULT_MLA_MODEL_NAME_FOR_TEST
cls.base_url = DEFAULT_URL_FOR_TEST
cls.process = popen_launch_server(
cls.model,
cls.base_url,
timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
other_args=[
"--trust-remote-code",
"--enable-hierarchical-cache",
]
+ hicache_args,
)
@classmethod
def tearDownClass(cls):
kill_process_tree(cls.process.pid)
def test_mmlu(self):
args = SimpleNamespace(
base_url=self.base_url,
model=self.model,
eval_name="mmlu",
num_examples=64,
num_threads=32,
)
metrics = run_eval(args)
self.assertGreater(metrics["score"], 0.5)
def test_mgsm_en(self):
args = SimpleNamespace(
base_url=self.base_url,
model=self.model,
eval_name="mgsm_en",
num_examples=None,
num_threads=1024,
)
metrics = run_eval(args)
self.assertGreater(metrics["score"], 0.8)
if __name__ == "__main__":
unittest.main()